Product Recommender System Based on Visually Similarity

Abstract

The need, for recommendation systems in the online shopping industry has become more evident due to the impact of the COVID 19 pandemic, which has shifted consumer behavior towards platforms. Conventional recommendation systems based on user activity data often struggle to capture consumer preferences regarding product visuals. This study explores the emerging field of recommendation systems utilizing similarity to enhance product suggestions. By examining Convolutional Neural Networks (CNN) and k Nearest Neighbors (KNN) in analyzing product images for similarities we aim to assess their influence on recommendation accuracy and personalization. A key contribution of this study is blending visual similarity metrics with recommendation algorithms showing that this combined approach significantly boosts user satisfaction through personalized and relevant product recommendations. Research findings indicate that integrating visual similarity can notably enhance user satisfaction by offering relevant product recommendations. Additionally we discuss how incorporating visual similarity metrics, into recommendation algorithms could transform e commerce by providing a tailored and immersive shopping experience. This analysis not provides insight into the status of visual based recommendation systems but also suggests future paths, for investigation and integration to improve online shopping platforms.